Home A proof of a conjecture by Jabara on groups in which all involutions are odd transpositions
Article Publicly Available

A proof of a conjecture by Jabara on groups in which all involutions are odd transpositions

  • Egle Bettio EMAIL logo
Published/Copyright: March 12, 2021

Abstract

In this paper, we prove that if 𝐺 is a group generated by elements of order two with the property that the product of any two such elements has order 1, 2, 3 or 5 with all possibilities occurring, then G A 5 or G PSU ( 3 , 4 ) . This provides an affirmative answer to Problem 19.36 in the Kourovka notebook.

1 Introduction

The main result in this paper is the following.

Theorem A

Let 𝐺 be a group, let Γ 2 be the set of elements of order 2 in 𝐺, and assume that

  1. G = Γ 2 ,

  2. if a , b Γ 2 , then the order of a b is 1, 2, 3 or 5 and all these values actually occur.

Then G A 5 or G PSU ( 3 , 4 ) .

Theorem A provides an affirmative answer to a conjecture by E. Jabara [13, Problem 19.36]. Furthermore, while Jabara’s conjecture refers to periodic groups, Theorem A applies to all groups.

Our proof is self-contained and requires only basic knowledge of group theory and some computations are carried out thanks to GAP software [16]. We also want to underline that, in principle, many computations with GAP could be verified by means of classical methods (in particular coset enumeration) which can be found described in the books [6, 12]. In any case, we have to admit that writing this article would have been impossible without the support of some software to test presentations of groups.

In finite group theory, hypotheses (i) and (ii) of Theorem A are well known, thanks to the seminal papers of B. Fischer [7] and M. Aschbacher [1, 2, 3, 4] (see the book [5]). Let 𝐺 be a finite group generated by a set Δ of involutions, where Δ is a union of conjugacy classes in 𝐺. Suppose that, for any two involutions a , b Δ that do not commute, the order of a b is odd. Then Δ is called a set of odd transpositions in 𝐺. Groups generated by a set of 3-transpositions were classified by Fischer, and this work led to the discovery of three new sporadic simple groups which are now called by the name of their discoverer. The most familiar example of groups investigated by Fischer is that of symmetric groups. Fischer’s approach to the study of finite groups is sometimes referred to as internal geometric analysis.

As F. Timmesfeld pointed out in [15, § 1], many of the above mentioned results can be extended to infinite groups if we introduce suitable finiteness conditions (precisely, he extrapolates from Gorenstein’s book [9]: “[…] the internal geometric analysis […] can reasonably be regarded as second in importance for the classification of finite simple groups only to local group-theoretic analysis. Now, probably contrary to local group-theory which depends on Sylow’s theorem, I always had the feeling that the internal geometric analysis is not restricted to finite groups. But dealing with abstract infinite groups it seems clear that one needs some finiteness condition to distinguish the classical and Lie-type groups over 𝑘 from the infinite dimensional groups.”). In the paper [11], Jabara and Zakavi use the internal geometric analysis in order to prove the local finiteness of some periodic groups generated by odd transpositions, thus obtaining a characterization of simple Suzuki groups Sz ( 2 2 m + 1 ) . Immediately after Jabara realized that, since in the group PSU ( 3 , 4 ) , the product of two involutions can have order 1, 2, 3 or 5 (with all possibilities occurring), this property made PSU ( 3 , 4 ) an ideal candidate to be characterized in the class of periodic groups, knowing only the order of the product of two of its involutions (assuming, however, the strong hypothesis that Δ = Γ 2 is the set of all involutions), and therefore he proposed [13, Problem 19.36].

In this article, we will provide a positive answer to Jabara’s conjecture, avoiding any further finiteness conditions.

2 Notation

Let 𝐺 be a group. If g G , we denote by o ( g ) the order of 𝑔, and if X G , we define the spectrum of 𝑋 as the set

ϖ ( X ) = { o ( x ) x X } .

We have ϖ ( G ) N { } , and 𝐺 is a periodic group if and only if ϖ ( G ) .

If n N , we define the subset of elements of order 𝑛 in 𝑋,

Γ n ( X ) = { x X o ( x ) = n } .

We shall write Γ n instead of Γ n ( G ) if there is no possible ambiguity; we remark that Γ n if and only if n ϖ ( G ) . Finally, if X , Y G , we define X Y to be the set { x y x X , y Y } and Π ( X ) = Γ 2 ( X ) Γ 2 ( X ) . Again, we shall write Π instead of Π ( G ) if there is no possible ambiguity.

If 𝐴 and 𝐵 are two groups, we denote by A . B (respectively A : B ) any extension (respectively split extension) of 𝐴 by 𝐵 which is not isomorphic to A × B .

We denote by C n the cyclic group of order 𝑛, by D 2 n the dihedral group of order 2 n , by S n and A n the symmetric and the alternating groups of degree 𝑛. Furthermore, we denote by L 150 = ( C 5 × C 5 ) : S 3 the group of order 150 such that L 150 is a Frobenius group of order 75. We will introduce additional notation throughout the paper; anything which remains unexplained can be found in [8] or [14].

Let 𝐺 be a group, and let Π = Π ( G ) . In the following, we will consider these hypotheses:

  1. G = Γ 2 ;

  2. ϖ ( Π ) = { 1 , 2 , 3 , 5 } .

Our main result can then be restated as follows.

Theorem A

Let 𝐺 be a group which satisfies hypotheses (H.1) and (H.2).

  1. If Γ 4 = , then G A 5 .

  2. If Γ 4 , then G PSU ( 3 , 4 ) .

Remark 2.1

Let 𝐺 be a group which satisfies hypotheses (H.1) and (H.2), and let 𝐻 be a subgroup of 𝐺. Then ϖ ( Π ( H ) ) { 1 , 2 , 3 , 5 } , but clearly, it can happen that H Γ 2 ( H ) (and also that Γ 2 ( H ) = ).

In principle, nothing can be said about quotients of 𝐺.

We summarize some of the properties of PSU ( 3 , 4 ) that are useful for our purposes.

Remark 2.2

Let G PSU ( 3 , 4 ) , and let 𝑆 be a Sylow 2-subgroup of 𝐺.

(A) Then 𝐺 has a unique conjugacy class of involutions, i.e. Γ 2 = a G ( a Γ 2 ).

(B) 𝑆 has order 2 6 and class 2, Γ 2 ( S ) = Z ( S ) C 2 × C 2 , and 𝑆 is a Suzuki 2-group in the sense of Higman [10]. Moreover, N G ( S ) S : C 15 , and its Hall { 2 , 3 } -subgroup L 192 = S : C 3 is a Frobenius group of order 192. The Frobenius group L 48 = ( C 4 × C 4 ) : C 3 (of order 48) is a subgroup of L 192 . (We want to point out that there are two non-isomorphic Frobenius groups of order 48 and five non-isomorphic Frobenius groups of order 192.)

(C) The maximal subgroups of 𝐺 are

C 13 : C 3 , L 150 ( C 5 × C 5 ) : S 3 , C 5 × A 5 , N G ( S ) S : C 15 .

(D) Let H , K G be such that H A 5 K and H K . Then H K = 1 or H K C 2 × C 2 . Moreover, if H K C 2 × C 2 , then

N H ( H K ) , N K ( H K ) ( C 4 × C 4 ) : C 3 = L 48

has the presentations a , x , y a 2 , x 3 , y 3 , ( a x ) 3 , ( a y ) 3 , a ( x y - 1 ) 2 , or, more simply,

( D ) L 48 x , y x 3 , y 3 , ( x y ) 3 , ( x y - 1 ) 4 .

(E) Let L 192 = S : C 3 , let L 48 x , y as in ( D ), and let a = ( x y - 1 ) 2 . If L 48 < L 192 , then there is some z Γ 3 ( L 192 ) such that L 192 has the presentation

( E ) x , y , z x 3 , y 3 , z 3 , ( x y ) 3 , ( y z ) 3 , ( z x ) 3 , ( x y - 1 ) 2 a , ( y z - 1 ) 2 a , ( z x - 1 ) 2 a x .

3 Preliminary results

From now, and throughout the paper, we will denote by 𝐺 a group that satisfies hypotheses (H.1) and (H.2). For every a Γ 2 , we write E a = Γ 2 ( C G ( a ) ) .

Lemma 3.1

Let a , b , c Γ 2 be such that [ a , b ] = 1 , [ b , c ] 1 , a b , a c . Then a , b , c A 5 and ( o ( b c ) , o ( c a ) , o ( a b c ) ) { ( 3 , 5 , 5 ) , ( 5 , 3 , 5 ) , ( 5 , 5 , 3 ) } .

Proof

Taking into account that a b Γ 2 , we can consider the groups

J = J ( , i 1 , i 2 ) = a , b , c a 2 , b 2 , c 2 , ( a b ) 2 , ( b c ) , ( c a ) i 1 , ( a b c ) i 2

with { 3 , 5 } and i 1 , i 2 { 2 , 3 , 5 } . A computation with GAP shows that

J ( 3 , 5 , 5 ) J ( 5 , 3 , 5 ) J ( 5 , 5 , 3 ) A 5 .

Moreover, J ( 5 , 5 , 5 ) PSL ( 2 , 11 ) , and this case must be excluded because D 12 is a subgroup of PSL ( 2 , 11 ) . For the other values of ℓ, i 1 and i 2 , the group 𝐽 is isomorphic to a group in the set { 1 , C 2 , C 2 × C 2 , D 6 , D 10 } . If J { 1 , C 2 , D 6 , D 10 } , then 𝐽 does not contain two distinct involutions which commute. If J C 2 × C 2 , then 𝐽 does not contain two distinct involutions which do not commute. In both cases, the hypotheses of the lemma are not satisfied, and therefore we obtain a contradiction. So the statement is proved. ∎

Lemma 3.2

𝐺 contains a unique conjugate class of involutions.

Proof

We begin by observing that Γ 2 ( Z ( G ) ) = : assume, arguing by contradiction, a Z ( G ) Γ 2 , and let b , c Γ 2 be such that o ( b c ) = 3 (or o ( b c ) = 5 ). Since a Z ( G ) , we have a b Z ( G ) , and ( a b ) c has order 6 (or 10), a contradiction.

Let a , b Γ 2 , a b . If [ a , b ] 1 , then a , b is isomorphic to D 6 or to D 10 , so 𝑏 is conjugate to 𝑎. If [ a , b ] = 1 , then (by (H.1)) we can choose c Γ 2 such that [ b , c ] 1 . From Lemma 3.1, a , b , c A 5 , and 𝑏 is conjugate to 𝑎 in a , b , c . ∎

Lemma 3.3

𝐺 has no subgroups isomorphic to C 2 × C 2 × C 2 . In particular, if a Γ 2 , then E a C 2 × C 2 and C G ( a ) N G ( E a ) .

Proof

Let a , b , c Γ 2 be such that a , b , c C 2 × C 2 × C 2 , and let t Γ 2 be such that a t Γ 3 . By Lemma 3.1, we obtain

a , b , t a , c , t a , b c , t A 5

and b t , c t , b c t Γ 5 . Since [ b , t ] 1 [ c , t ] , we have also b , c , t A 5 , and hence b t or c t or b c t has order 3. Assume b t Γ 3 . Then ( b t ) 5 = 1 = ( b t ) 3 implies b t = 1 and t = b , a contradiction.

Hence if b , c C G ( a ) Γ 2 , then [ b , c ] 1 . But

a , b , c a 2 , b 2 , c 2 , ( a b ) 2 , ( c a ) 2 , ( b c ) D 4 , where { 3 , 5 } ,

which is impossible. So C G ( a ) Γ 2 = { a , b , a b } and E a = a , b C 2 × C 2 .

Let g C G ( a ) , and suppose E a = a , b . Then we have [ a , b g ] = 1 , b g E a and g N G ( E a ) . ∎

Lemma 3.4

Let H G be such that H A 5 or H PSU ( 3 , 4 ) . If a Γ 2 ( H ) , then E a H .

Proof

The claim is an easy consequence of Lemma 3.3. ∎

Lemma 3.5

Let a , b Γ 2 be such that [ a , b ] 1 . Then E a , E b A 5 .

Proof

Let c E b with 1 c b . Then, by Lemma 3.1, H = a , b , c A 5 . By Lemma 3.4, we have E a H and E b H , and hence we can conclude that E a , E b A 5 . ∎

Lemma 3.6

Let 𝑇 be a group such that ϖ ( Π ( T ) ) { 1 , 2 , 3 , 5 } , let 𝑁 be a normal subgroup of 𝑇, and assume that 𝑁 is periodic and Γ 2 ( N ) = . Then we have ϖ ( Π ( T / N ) ) = ϖ ( Π ( T ) ) ; in particular, ϖ ( Π ( T / N ) ) { 1 , 2 , 3 , 5 } .

Proof

Let a N , b N Γ 2 ( T / N ) . Since 𝑁 is periodic and without involution, we can assume a , b Γ 2 ( T ) and ( a b ) k = 1 for some k { 2 , 3 , 5 } . Then we have ( a N b N ) k = ( a b N ) k = N . ∎

Let L 18 ( C 3 × C 3 ) : C 2 be the Frobenius group of order 18, let E 27 be the exstraspecial group of order 27 and exponent 3, and let W 54 = E 27 : C 2 be such that W 54 / Z ( W 54 ) L 18 . If ρ = { a 2 , b 2 , c 2 , ( a b ) 3 , ( b c ) 3 , ( c a ) 3 } , then presentations for L 18 and W 54 are

(3.1) L 18 = a , b , c ρ L , where ρ L = ρ { ( a b c ) 2 } ,
(3.2) W 54 = a , b , c ρ W , where ρ W = ρ { ( a b c ) 3 }
( ρ W implies that ( b c a ) 3 = 1 = ( c a b ) 3 ), and in (3.2), we have

Z ( W 54 ) = ( a b c ) 2 C 3 .

Lemma 3.7

𝐺 does not contain subgroups isomorphic to L 18 or to W 54 .

Proof

We argue by contradiction. Assume that L 18 < G , and let a , b , c Γ 2 be such that L 18 a , b , c ρ L is as in (3.1). Let t Γ 2 { a } be such that [ a , t ] = 1 , and let

Δ 0 = { b , c , b c , a b c } Γ 2 ( L 18 ) { a } .

For every δ Δ 0 , we have a δ Γ 3 , and hence, by Lemma 3.1, t δ , a t δ Γ 5 . Let

σ = { t 2 , ( a t ) 2 , ( b t ) 5 , ( c t ) 5 , ( b c t ) 5 , ( a b c t ) 5 , ( a t b ) 5 , ( a t c ) 5 } ,

and define

J ( ) = a , b , c , t ρ L σ { ( b t c ) } , { 2 , 3 , 5 } .

A computation with GAP shows that J ( 2 ) = J ( 3 ) = J ( 5 ) = 1 , a contradiction.

Assume that W 54 < G , and let a , b , c Γ 2 be such that W 54 = a , b , c is as in (3.2). Let t Γ 2 { a } be such that [ a , t ] = 1 , and let z = ( a b c ) 2 Z ( W ) . By Lemma 3.3, t z E a , so t z = t or t z = a t , and since 𝑧 has order 3, we must have t z = t , so z Z ( a , b , c , t ) .

Let S ¯ = a ¯ , b ¯ , c ¯ , t ¯ = a , b , c , t / z , R ¯ = a ¯ , b ¯ , c ¯ L 18 , and let 𝑆 and 𝑅 be the complete preimage of R ¯ and S ¯ in 𝐺. By Lemma 3.6, ϖ ( Π ( S ¯ ) ) { 1 , 2 , 3 , 5 } , and since a t b Γ 5 ( Π ( S ¯ ) ) , ϖ ( Π ( S ¯ ) ) = { 1 , 2 , 3 , 5 } . So S ¯ satisfies hypotheses (H.1) and (H.2) and R ¯ < S ¯ with R ¯ L 18 . This is in contradiction with what we have shown in the first part of the proof. ∎

Lemma 3.8

Let x Π Γ 3 . Assume that x = a b with a , b Γ 2 . Then we have { t Γ 2 x t = x - 1 } = { a , b , a b } .

Proof

Let t Γ 2 be such that x t = x - 1 , assume t { a , b , a b } , and let

J ( ) = a , b , t a 2 , b 2 , t 2 , ( a b ) 3 , ( a b t ) 2 , ( b t ) , { 2 , 3 , 5 } .

An easy computation shows that J ( 2 ) D 12 and J ( 5 ) D 30 , and we must rule out such cases by (H.2), while J ( 3 ) ( C 3 × C 3 ) : C 2 = L 18 , against Lemma 3.7. ∎

4 Subgroups generated by three involutions

Definition 4.1

Let

a , b , c Γ 2 with o ( a b ) o ( b c ) o ( c a ) .

If the subgroup H = a , b , c cannot be generated by less than three involutions, then 𝐻 is said to be a subgroup of type ( o ( a b ) , o ( b c ) , o ( c a ) ) .

In Lemma 3.1, we have shown that if a group 𝐻 is of type ( 2 , i , j ) , then H A 5 , i { 3 , 5 } and j = 5 . In this section, we will determine the structure of subgroups of 𝐺 of type ( , i , j ) with 3 .

Lemma 4.2

If H G is of type ( 3 , 3 , 3 ) , then H L 150 .

Proof

Let

J = J ( i 1 , i 2 ) = a , b , c a 2 , b 2 , c 2 , ( a b ) 3 , ( b c ) 3 , ( c a ) 3 , ( b a c ) i 1 , ( c b c a ) i 2

with i 1 , i 2 { 2 , 3 , 5 } . A computation shows that J ( 2 , 3 ) S 4 , which is impossible, J ( 3 , 3 ) E 27 : C 2 , which is excluded for Lemma 3.7 and J ( 5 , 3 ) L 150 . Since for other values of i 1 and i 2 , we have | J | 6 , the lemma is proved. ∎

Lemma 4.3

If H < G is of type ( 3 , 3 , 5 ) , then 𝐻 can be isomorphic only to A 5 , to L 150 or to PSU ( 3 , 4 ) .

Proof

Let

J = J ( i 1 , i 2 ) = a , b , c a 2 , b 2 , c 2 , ( a b ) 3 , ( b c ) 3 , ( c a ) 5 , ( b a c ) i 1 , ( c b c a ) i 2

with i 1 , i 2 { 2 , 3 , 5 } .

A computation shows that J ( 5 , 2 ) C 2 × A 5 with Z ( J ) = ( a b c ) 3 , so in this case, we have a , b , c A 5 , J ( 3 , 3 ) L 150 and J ( 5 , 5 ) C 2 × PSU ( 3 , 4 ) with Z ( J ) = ( a b c ) 15 , so in this case, a , b , c PSU ( 3 , 4 ) . Since for other values of i 1 and i 2 , we have | J | 2 , the lemma is proved. ∎

Lemma 4.4

If H < G is such that H L 150 , then for every a Γ 2 ( H ) , we have H , E a PSU ( 3 , 4 ) .

Proof

Let H = a , b , c ρ L 150 , where

ρ = { a 2 , b 2 , c 2 , ( a b ) 3 , ( b c ) 3 , ( c a ) 3 , ( a b c ) 5 } .

Then w = ( c a b ) 2 Γ 5 ( H ) and [ a , w ] = 1 . Let t Γ 2 be such that E a = a , t . By Lemma 3.3, t w E a , and since 𝑤 has odd order, [ t , w ] = 1 .

If h = b or h = c , then, by Lemma 3.1, we have t h , a t h Γ 5 . Since a b c Γ 5 again by Lemma 3.1, we have

  1. t b c Γ 3 and a t b c Γ 5 or

  2. t b c Γ 5 and a t b c Γ 3 .

Let σ = { t 2 , ( a t ) 2 , [ t , w ] , ( t b ) 5 , ( a t b ) 5 , ( t c ) 5 , ( a t c ) 5 } , and consider the groups

J ( ) = a , b , c , t ρ , σ , ( t b c ) , ( a t b c ) 8 - , where { 3 , 5 } .

A computation with GAP shows that J ( 3 ) PSU ( 3 , 4 ) J ( 5 ) . ∎

Definition 4.5

Three involutions a , b , c Γ 2 are said to be independent if

E a E b , E c = E b E c , E a = E c E a , E b = 1 .

We remark that, if a , b , c are independent, then E a E b E c E a , and hence, by Lemma 3.5, E a , E b E b , E c E c , E a A 5 .

Lemma 4.6

Let a , b , c Γ 2 be three independent involutions of 𝐺. Then we have E a , E b , E c PSU ( 3 , 4 ) .

Proof

Let E a = a , α , E b = b , β and E c = c , γ , and let J = E a , E b , E c . Fix 𝑎. Then, by Lemma 3.1, there are b 1 E b and c 1 E c such that a b 1 , a c 1 Γ 3 . We can, without loss of generality, assume that b = b 1 and c = c 1 , and hence H = a , b , c is of type ( 3 , 3 , 3 ) or ( 3 , 3 , 5 ) . From Lemmas 4.2 and 4.3, we must consider the following three cases.

(1) H L 150 . If J 1 = H , E a , J 2 = H , E b and J 3 = H , E a , then, by Lemma 4.4, we can conclude that J 1 J 2 J 3 PSU ( 3 , 4 ) and, by Lemma 3.4, E a E b E c J 1 J 2 J 3 .

Hence J = J 1 = J 2 = J 3 PSU ( 3 , 4 ) .

(2) H A 5 , by Lemma 3.5, H = E a , E b , and by Lemma 3.4, E c H , against the hypotheses. So this case is not possible.

(3) H PSU ( 3 , 4 ) , and by Lemma 3.4, E a , E b , E c H .

Hence J = H PSU ( 3 , 4 ) . ∎

An immediate consequence of Lemma 3.1 and Lemma 4.6 is that, for every a , b , c Γ 2 , the subgroup a , b , c is finite.

Lemma 4.7

Let x Γ 3 . Then there is an element d Γ 2 such that x d = x - 1 or x Z ( G ) .

Proof

If [ a , x ] = 1 for every a Γ 2 , then we should have [ x , Γ 2 ] = 1 and x Z ( G ) , since G = Γ 2 . If [ a , a x ] = 1 for every a Γ 2 , then a , x A 4 and ( a x ) 3 = 1 for every a Γ 2 . Let b Γ 2 be such that a b Γ 3 . Taking into account that a x b Γ 5 , with GAP, we can verify that the group

H = a , b , x a 2 , b 2 , x 3 , ( a x ) 3 , ( b x ) 3 , ( a b x ) 3 , ( a b ) 3 , ( a x b ) 5

has order 3 (and, in 𝐻, a = b ), which is impossible.

Let a Γ 2 be such that [ a , x ] 1 , let b = a x , c = b x and J = E a , E b , E c . By Lemma 4.6, 𝐽 is finite, and since we can easily verify that E a x = E b and E b x = E c , we have J = J x . Since E a x E c , we must consider the following two cases.

(1) J A 5 . Since Aut ( J ) A 5 : C 2 , we have x J , and we conclude because every element of Γ 3 ( A 5 ) is inverted by some involution in A 5 .

(2) J PSU ( 3 , 4 ) . Since Aut ( J ) PSU ( 3 , 4 ) : C 4 , we have x J , and we conclude because every element of Γ 3 ( PSU ( 3 , 4 ) ) is inverted by some involution in PSU ( 3 , 4 ) . ∎

5 Proof of Jabara’s conjecture

Proposition 5.1

If Γ 4 = , then G A 5 .

Proof

Let H G with H A 5 , and let a , b Γ 2 ( H ) be such that E a E b . From Lemma 3.5, we have H = E a , E b . Let c Γ 2 be such that c H . Then, by Lemma 3.4, H E c = 1 , and by Lemma 4.6, E a , E b , E c PSU ( 3 , 4 ) .

Since PSU ( 3 , 4 ) contains elements of order 4, this forces Γ 2 H , and by (H.1), G = Γ 2 , so G = H A 5 . ∎

Lemma 5.2

If Γ 4 , then there is H G such that H PSU ( 3 , 4 ) .

Proof

Let a , b Γ 2 with a b b a . Then E a , E b A 5 . Since Γ 4 ( A 5 ) = and G = Γ 2 , there is c Γ 2 such that c E a , E b . From Lemma 4.6, we have H = E a , E b , E c PSU ( 3 , 4 ) . ∎

The (Frobenius) groups L 48 and L 192 are defined in Remark 2.2 (B).

Lemma 5.3

Let E a G and x , y Γ 3 be such that E a , x A 4 E a , y . If E a , x E a , y , then

  1. E a , x , y = x , y ( C 4 × C 4 ) : C 3 = L 48 , and we have

    (5.1) x , y = a , x , y x 3 , y 3 , ( x y ) 3 , ( x y - 1 ) 4

    or

    (5.2) x , y = a , x , y x 3 , y 3 , ( x y ) 4 , ( x y - 1 ) 3 .

    In particular, we can choose y * Γ 3 ( L 48 ) so that a = ( x y * - 1 ) 2 .

  2. There is only one H G such that H PSU ( 3 , 4 ) and E a , x , y < H .

Proof

Since x , y Z ( G ) , by Lemma 4.7, we have that there are b , c Γ 2 such that x b = x - 1 and y c = y - 1 .

Let X = E a , E b A 5 and β = b x Γ 2 . Let a 1 E a be such that a 1 b Γ 3 , and let a 2 = a 1 x . Then E a = E a 1 and X = E a 1 , E b = a 1 , a 2 , b ξ , where

ξ = { a 1 2 , a 2 2 , b 2 , ( a 1 a 2 ) 2 , ( a 1 b ) 3 , ( a 2 b ) 5 , ( a 1 a 2 b ) 5 } .

Let

T = a 1 , a 2 , b , x ξ { x 3 , a 1 x a 2 , ( a 1 x ) 3 , ( b x ) 2 } .

A computation shows that T A 5 , and hence T = X , E a , E b , E β = E a , E b , so E β X and x X . In the same way, we get that y Y = E a , E c .

Let H = X , Y = E a , E b , E c . If H A 5 , then E a , x = E a , y , which is excluded from the hypotheses, so by Lemma 4.6, H PSU ( 3 , 4 ) . The presentation (5.1) for E a , x , y follows from ( D ) in Remark2.2, and we obtain (5.2) by replacing 𝑦 with y - 1 .

The uniqueness of 𝐻 follows from Lemma 3.8. ∎

Lemma 5.4

Let E a G and x , y , z Γ 3 be such that x , y , z N G ( E a ) and E a , x , y E a , y , z L 48 . If z E a , x , y , then

  1. E a , z , x L 48 ,

  2. E a , x , y , z = x , y , z S : C 3 = L 192 ,

  3. if x , y and 𝑧 are chosen in such a way that ( x y - 1 ) 2 = a = ( y z - 1 ) 2 , then ( z x - 1 ) 2 { a x , a a x } . If ( z x - 1 ) 2 = a x , then a presentation for x , y , z is ( E ) in Remark 2.2.

Proof

For every w Γ 3 ( L 48 ) and every a Γ 2 ( L 48 ) , we have

E a , w = a , w A 4 .

In particular, E a , x A 4 E a , z and, by Lemma 5.3 (a),

E a , x , z = x , z L 48 .

From Lemma 5.3 (a), we can suppose, without loss of generality, that 𝑦 is such that ( x y - 1 ) 2 = a , ( x y ) 3 = 1 and 𝑧 is such that ( y z - 1 ) 2 = a , ( y z ) 3 = 1 . In particular, the following set of relations hold in 𝐺:

ρ = { x 3 , y 3 , z 3 , ( x y ) 3 , ( y z ) 3 , ( x y - 1 ) 2 a , ( y z - 1 ) 2 a } .

Since x , y , z ρ { ( z x ) 4 , ( z x - 1 ) 3 , ( y z ) 2 b } A 5 , for every b E a , we have z x Γ 3 and ( z x - 1 ) 2 = c { a , a x , a a x } .

Since x y , z Γ 3 and E a , x y A 4 , by Lemma 5.2, E a , x y , z L 48 . We remark that

x , y , z ρ { ( y z ) 3 , ( y z - 1 ) 4 , ( x y z ) 3 } A 4 ,

so we must have x y z Γ 4 and ( x y z ) 2 = d { a , a x , a a x } .

Consider the groups

J = J ( c , d ) = x , y , z ρ { ( x y ) 3 , ( y z - 1 ) 2 c , ( x y z ) 2 d } , c , d { a , a x , a a x } .

A computation with GAP shows that J ( a x , a a x ) L 192 J ( a a x , a x ) and that, if { c , d } { a x , a a x } , then J ( c , d ) C 2 4 : C 3 , which is impossible since L 48 J and Γ 4 ( L 48 ) .

We can verify that a presentation for J ( a x , a a x ) is ( E ). ∎

Lemma 5.5

Let E a G and x , y , z , w Γ 3 be such that x , y , z , w N G ( E a ) . If E a , x , y , z L 192 , then w E a , x , y , z .

Proof

We argue by contradiction assuming that w E a , x , y , z .

Clearly, we can suppose E a , x , y , w E a , y , z , w E a , z , x , w L 192 and that x , y , z has the presentation ( E ). In particular,

( x y - 1 ) 2 = a = ( y z - 1 ) 2 , ( z x - 1 ) 2 = a x , ( x y z ) 2 = a a x

(see the proof of Lemma 5.4), and since E a , x , y , w L 192 , we can choose 𝑤 such that ( y w - 1 ) 2 = a and ( w x - 1 ) 2 = a x . Hence the set ρ = ρ 1 ρ 2 ρ 3 of relations hold in 𝐺, where

ρ 1 = { x 3 , y 3 , z 3 , w 3 , ( x y ) 3 , ( x z ) 3 , ( x w ) 3 , ( y z ) 3 , ( y w ) 3 , ( z w ) 3 } ,
ρ 2 = { ( x y - 1 ) 2 a , ( y z - 1 ) 2 a , ( z x - 1 ) 2 a x , ( y w - 1 ) 2 a , ( w x - 1 ) 2 a x } ,
ρ 3 = { ( x y z ) 2 a a x , ( x y w ) 2 a a x , ( x z w ) 4 , ( y z w ) 4 } , where a = ( x y - 1 ) 2 .
Moreover, ( z w - 1 ) 4 = 1 , and there are b , c { a , a x , a a x } such that ( x z w ) 2 = b and ( y z w ) 2 = c . A computation with GAP shows that

J ( b , c ) = x , y , z , w ρ { ( z w - 1 ) 4 , ( x z w ) 2 b , ( y z w ) 2 c } C 2 6 : C 3

for every choice of b , c E a , a contradiction, and hence we must conclude that w x , y , z . ∎

Proposition 5.6

If Γ 4 , then G = PSU ( 3 , 4 ) .

Proof

Let H = E a , E b , E c PSU ( 3 , 4 ) with a , b , c independent involutions of 𝐺. If H < G , then there is d Γ 2 such that d H and, by Lemma 3.4, H E d = 1 . By Lemma 4.6, K = E a , E b , E d PSU ( 3 , 4 ) , by Lemma 3.5, D = E a , E b A 5 , and obviously, D H K .

Let 𝑥 be an element of order 3 in N D ( E a ) , and let y , z N H ( E a ) Γ 3 be such that x , y , z L 192 . If w N K ( E a ) Γ 3 is such that E a , x , w L 48 , then, by Lemma 5.5, w x , y , z , and hence L 48 E a , x , w H K .

By Lemma 5.3 (b), there is a unique subgroup of 𝐺 isomorphic to PSU ( 3 , 4 ) which contains E a , x , w , and hence H = K . This forces Γ 2 H , and we can conclude, since H = Γ 2 = G . ∎

By gluing together Propositions 5.1 and 5.6, we obtain the proof of Theorem A and therefore of Theorem A.


Dedicated to Professor Enrico Jabara on the occasion of his 64th birthday


Acknowledgements

The author deeply thanks the referee for reading the paper very carefully and giving a lot of valuable suggestions kindly and patiently.

  1. Communicated by: Evgeny Vdovin

References

[1] M. Aschbacher, On finite groups generated by odd transpositions. I, Math. Z. 127 (1972), 45–56. 10.1016/S0304-0208(08)71819-XSearch in Google Scholar

[2] M. Aschbacher, On finite groups generated by odd transpositions. II, J. Algebra 26 (1973), 451–459. 10.1016/0021-8693(73)90007-0Search in Google Scholar

[3] M. Aschbacher, On finite groups generated by odd transpositions. III, J. Algebra 26 (1973), 460–478. 10.1016/0021-8693(73)90008-2Search in Google Scholar

[4] M. Aschbacher, On finite groups generated by odd transpositions. IV, J. Algebra 26 (1973), 479–491. 10.1016/0021-8693(73)90009-4Search in Google Scholar

[5] M. Aschbacher, 3-Transposition Groups, Cambridge Tracts in Math. 124, Cambridge University, Cambridge, 1997. Search in Google Scholar

[6] H. S. M. Coxeter and W. O. J. Moser, Generators and Relations for Discrete Groups, 4th ed., Ergeb. Math. Grenzgeb. (3) 14, Springer, Berlin, 1980. 10.1007/978-3-662-21943-0Search in Google Scholar

[7] B. Fischer, Finite groups generated by 3-transpositions. I, Invent. Math. 13 (1971), 232–246. 10.1007/BF01404633Search in Google Scholar

[8] D. Gorenstein, Finite Groups, 2nd ed., Chelsea, New York, 1980. Search in Google Scholar

[9] D. Gorenstein, Finite Simple Groups, Univ. Ser. Math., Plenum, New York, 1982. 10.1007/978-1-4684-8497-7Search in Google Scholar

[10] G. Higman, Suzuki 2-groups, Illinois J. Math. 7 (1963), 79–96. 10.1215/ijm/1255637483Search in Google Scholar

[11] E. Jabara and A. Zakavi, Periodic groups in which all involutions are odd transpositions (in Russian), Sibirsk. Mat. Zh. 60 (2019), no. 1, 229–237; translation in Sib. Math. J. 60 (2019), no. 1, 178–184. 10.33048/smzh.2019.60.119Search in Google Scholar

[12] D. L. Johnson, Presentations of Groups, 2nd ed., London Math. Soc. Stud. Texts 15, Cambridge University, Cambridge, 1997. 10.1017/CBO9781139168410Search in Google Scholar

[13] E. I. Khukhro and V. D. Mazurov, The Kourovka Notebook, Sobolev Institute of Mathematics, Novosibirsk, 2018. Search in Google Scholar

[14] D. J. S. Robinson, A Course in the Theory of Groups, 2nd ed., Grad. Texts in Math. 80, Springer, New York, 1996. 10.1007/978-1-4419-8594-1Search in Google Scholar

[15] F. G. Timmesfeld, Groups generated by 𝑘-transvections, Invent. Math. 100 (1990), no. 1, 167–206. 10.1007/BF01231184Search in Google Scholar

[16] The GAP: groups, algorithms, and programming, vers. 4.10.2 (2019), http://www.gap-system.org. Search in Google Scholar

Received: 2020-08-22
Revised: 2021-02-17
Published Online: 2021-03-12
Published in Print: 2021-09-01

© 2021 Walter de Gruyter GmbH, Berlin/Boston

Downloaded on 24.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/jgth-2020-0129/html
Scroll to top button